Dynamic lightpath establishment for service differentiation based on optimal MDP policy in all-optical networks with wavelength conversion

In this paper, we propose a dynamic lightpath establishment method for service differentiation in all-optical WDM networks with the capability of full-range, wavelength conversion. In the proposed method, first, the optimal lightpath establishment policy for the service differentiation is derive based on Markov decision process (MDP). The optimal policy considered here is to provide the service differentiation for the lightpath establishment, keeping the probability that the lightpath establishment fails small. Then, each node processes signaling messages according to this optimal policy. In this method, each node is required to store additional information about the number of established lightpaths and the action for the lightpath establishment. In numerical examples, we show that our proposed method can provide the service differentiation for the blocking probability that the lightpath establishment fails, decreasing the overall blocking probability. We also show the effectiveness of the proposed method in a uni-directional ring network.